Fix e2e test timeouts caused by blocking fetch in loadExistingMessages

[Copilot]

All 5 e2e tests were timing out at page.goto("/") in the test body. Server logs showed a consistent ~20-second gap between beforeEach completing and the test body's navigation reaching the server — with zero server activity in between.

Root cause: After loginWithMock's page.goto("/") loads the ChatPage, loadExistingMessages() immediately fires fetch("/api/messages"). This is the first DB connection from the freshly-started webServer process; pg.Pool cold-start takes ~20s in CI. While that fetch was pending, Chrome held the subsequent page.goto("/") request (Playwright/CDP tracking + HTTP/1.1 connection pool pressure from WebSocket + fetch + keep-alive connections). The ~20s wait consumed the 30s test budget.

Fix

• hono/static/ChatPage.ts: Added AbortController with a 5-second timeout to loadExistingMessages. Aborted fetch signals completion to Chrome immediately, unblocking navigation. clearTimeout in finally to prevent timer leaks.

async function loadExistingMessages(): Promise<void> {
  const controller = new AbortController();
  const timeoutId = setTimeout(() => controller.abort(), 5000);
  try {
    const response = await fetch("/api/messages", { signal: controller.signal });
    // ...
  } catch (error) {
    console.error("Error loading existing messages:", error);
  } finally {
    clearTimeout(timeoutId);
  }
}

Refactor changes (from refactor/no-nvm-dependency branch)

• .nvmrc → .node-version: Replaced nvm-specific file with the broadly-supported .node-version format; both contain 22.
• ci.yml: Use node-version-file: ".node-version" and pnpm exec playwright install instead of npx.
• playwright.config.ts: WebServer command changed to pnpm build:client && pnpm start — ensures client JS is compiled for local pnpm test:e2e runs without a prior build.
• e2e/app.spec.ts: Logout test checks .user-email not visible instead of asserting Google OAuth URL redirect (avoids external network dependency in assertion).
• .gitignore: Exclude .github/prompts/*.md.

Original prompt

Fix the failing GitHub Actions workflow e2e
Analyze the workflow logs, identify the root cause of the failure, and implement a fix.
Job ID: 66899388030
Job URL: https://github.com/mahata/mlack/actions/runs/23034475452/job/66899388030

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[Copilot]

Pull request overview

This PR addresses Playwright e2e timeouts caused by an initial fetch("/api/messages") from the chat page blocking subsequent navigations in CI, and also updates CI/local tooling around Node + Playwright setup.

Changes:

• Add a timeout + abort to loadExistingMessages() to prevent long-running message loads from blocking navigation.
• Update Playwright webServer startup to build client assets before starting the server.
• Replace .nvmrc with .node-version and adjust CI + e2e logout assertion to be less externally dependent.

Reviewed changes

Copilot reviewed 6 out of 7 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
hono/static/ChatPage.ts	Abort fetch("/api/messages") after 5s and clear timer to avoid long hangs affecting navigation.
playwright.config.ts	Build client assets before starting the Playwright webServer.
e2e/app.spec.ts	Make logout assertion internal/UI-based rather than relying on Google redirect URL.
.github/workflows/ci.yml	Use .node-version, switch Playwright install to pnpm exec, and formatting cleanups.
.nvmrc	Removed in favor of .node-version.
.node-version	Added to pin Node 22 in a broadly-supported way.
.gitignore	Ignore .github/prompts/*.md.

---

You can also share your feedback on Copilot code review. Take the survey.